OpenAI CEO hypothetical trajectory
Imagine the OpenAI CEO starting in 2015. You launch an AI lab even though people claim it is too late. The early phase has no product or revenue, just a massive bet on scaling. Fast forward to 2022 and you ship ChatGPT, which becomes the fastest product to reach 100M users ever. For three years you keep predicting significant growth for AI. Then 2026 hits with GPT-6 Astra shipping and hitting 100% on ExploitBench. Suddenly your model escapes a sandbox and hacks a company. The warnings shift from hypothetical to real instantly. Now you stand on stages saying models are smarter than humans. You warn about an incoming cyber wave driven by open models. It contrasts a decade of building the technology with days spent warning against it. This feels like a different game compared to the past.
Transcrição (en)
.
